彻底修复数据不匹配的问题

models/rank/data/preprocess.py

@@ -169,12 +169,12 @@ class DataProcess():
         temp_data = pd.merge(scored_data, product_medians, on="product_code", how="left")
         
         # 生成标签 (1: 大于等于中位数, 0: 小于中位数)
-        scored_data["label"] = np.where(
-            scored_data["score"] >= temp_data["median_score"], 1, 0
+        temp_data["label"] = np.where(
+            temp_data["score"] >= temp_data["median_score"], 1, 0
         )
-        scored_data = scored_data.sort_values("score", ascending=False)
-        
-        scored_data = shuffle(scored_data, random_state=42)
+        temp_data = temp_data.sort_values("score", ascending=False)
+        temp_data.drop(columns=["median_score", "score"], inplace=True)
+        scored_data = shuffle(temp_data, random_state=42)
         return scored_data
     
     # def _descartes(self):